# Computational modeling of language impairment and control in bilingual individuals with post-stroke aphasia and neurodegenerative disorders

> **NIH NIH R01** · BOSTON UNIVERSITY (CHARLES RIVER CAMPUS) · 2024 · $653,131

## Abstract

Per the 2020 US Census (data.census.gov), there are approximately 62 million Hispanic/Latino individuals living
in the US, and of these 15% are individuals aged 55 years and older 2. The US Census also projects that by
2030, this Hispanic population will increase to 74 million and by this time the number of older adults will
outnumber children. A separate but concerning statistic is that the WHO predicts a rising incidence of dementia
(78 million individuals by 2030) 3 and strokes (70 million survivors by 2030) 4 suggesting that there is an increased
urgency to provide clinical services for these populations. In order to do that, it is necessary to understand the
interaction between bilingualism (in Spanish-English speaking Hispanic individuals) and
neurological/neurodegenerative disorders. The problem is that bilingual speakers vary widely in how effectively
they process their two languages and how these processes may break down in neurological disorders. Thus, to
fully understand the nature of bilingual impairment (in stroke) and decline (in dementia), it would be necessary
to conduct prohibitively large-scale cross-sectional and longitudinal examinations of hundreds of bilingual
individuals with varying degrees of proficiency to accurately capture the variation in bilingual speakers.
Our central hypothesis is that computational simulations of bilingual language processing (in healthy aging),
language impairment and recovery (in stroke), and decline (in neurodegenerative disorders) is a powerful
approach in lieu of such studies. Computational modeling makes it possible to study an adult bilingual language
system that can vary by any language combination and proficiency at any single time point and characterize
change over time. BILEX, our computational model for bilingual language processing, has an already-proven
ability to simulate bilingual post-stroke aphasia (BPSA) and bilingual semantic dementia (BSD) as well as
rehabilitation outcomes for post-stroke individuals. Computational simulations can, thus, be used to effectively
represent not just known patient cases but also generalize to cases for which we do not yet have any patient
data. Consequently, our specific aims are to explain how different types of observed impairments in BPSA and
BSD arise. For that reason, we need to first characterize these impairments in more detail, and understand how
they may arise using a computational structure of maps and connections (Aim 1). Armed with such an
understanding, the second aim is to extend them over time, i.e. to explain how these mechanisms result in
recovery after a stroke (in aphasia) and decline (in dementia) (Aim 2). The third aim, then, is to understand how
these processes interact with language selection and control, by including subcortical conditional routing
mechanisms to our BILEX model (Aim 3). Each aim allows for progressively more detailed characterizations and
explanations of modeling impairment and recovery in bilingual post-st...

## Key facts

- **NIH application ID:** 10920366
- **Project number:** 5R01DC020653-02
- **Recipient organization:** BOSTON UNIVERSITY (CHARLES RIVER CAMPUS)
- **Principal Investigator:** Swathi Kiran
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $653,131
- **Award type:** 5
- **Project period:** 2023-09-04 → 2028-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10920366

## Citation

> US National Institutes of Health, RePORTER application 10920366, Computational modeling of language impairment and control in bilingual individuals with post-stroke aphasia and neurodegenerative disorders (5R01DC020653-02). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10920366. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*